Hi, I’ve been frustrated lately trying to put all of my configuration into one simple clonable. The issue is many nix packages are not properly declaratively configurable. However the HM implementation of them is. Many NixOS users are effectively only using one user space, but are pushed in HM to declaratively configure. A fork is a great opportunity to expand on the configuration of packages and allow a Home Manager implemention to focus on managing user home’s. Obviously this would be a very significant undertaking, but it can be something strived for.
12 Likes
Thank you so much for posting this!
I’d actually been mulling over this exact thing the last couple of hours. 
2 Likes
It would seem to make sense to have a Home Manager SIG which collaborates more directly with HM than is the case with nix?
4 Likes
I completely agree. I have been thinking about this over the last few days as well. I think we can leave such an effort until the initial roadmap is completed, but a first-party solution that more cleanly integrates and interoperates with the os project would be excellent.
6 Likes
I’d love to see HM just be integrated into the project like how Guix handles it.
3 Likes
I think if we can integrate home-manager into the project, it would be nice to be able to self bundle its configuration, instead of placing symlink in ~/.config
We can learn from this and try something similar
Fork is a great opportunity to make breaking changes
4 Likes
These are great suggestions. Sounds like we have some Home Manager SIG members handy here?
I think, rather than focusing on home management with configuration, making packages that are more declaratively configurable divorce with whatever Home Manager implementation, then the HM implementation uses those packages is a better approach. I don’t see why user space configuration and system wide configuration options couldn’t just be the same. Of course, a different approach to home management could also be good, my point was just dropping the differences in configuration though.
1 Like
I think it may be a good idea to hold off on non-critical things like home-manager support until later. I think after we’ve gotten things up and running, completing our roadmap, then we can definitely spin up initiatives like this.
3 Likes
Even the planning efforts? I’m having a hard time seeing the downside of making the most of enthusiasm at this point, with a sensible limit on what can move from planning & testing to execution at a given phase.
Enthusiasm is not endless or a strictly controllable resource. At the point you go “OK - now I’m ready, get hyped now pls” the best of it may well have dissipated, leaving you with merely competent output vs. high-grade inspiratio-splotion 
1 Like
We can create a SIG focused on home-manager things. I think I’m more concerned that we should focus on the roadmap right now which requires more work done on nixpkgs and nix than supplementary projects/features. Maybe we create the group now so people can be organized for the future, but understand that it’s probably best to wait on spending a bunch of time on it?
1 Like
I would recommend creating the group, letting people go ham on the problem, knowing the project is not in a state where things can move out of planning & testing, while continuing to clearly communicate roadmap priorities.
Even when people are paid employees, it is rarely conducive to productivity to discourage useful productivity too heavily - whether it is convenient or not.
5 Likes
Just saw this myself, I’m curious but knowing the nix foundation it likely won’t go anywhere.
I’m cautiously optimistic